Notes

Questions

DIET BREADTH WOULD BE INTERESTING TO LOOK AT (how many different things each lizard ate)

Plan of analytical attack

1. How does invert community vary with land use and veg cover?

• Visualize with NMDS
  • Point for each trap (color-coded by pitfall vs. sticky trap type)
  • Vectors for veg cover (by life form)
    • We have trap-specific vegetation data
  • Vector for land use (might need to be binary –> polygon)
• Model with multivariate GLMM (via mvabund)

taxa abundances ~ land use + veg cover (by lifeform) + RE:site + RE:trap type

• How to combine pitfall/sticky data?
  
• Option 1: As described above (for NMDS, color-code by trap type; for GLMM, include trap type as a RE)
  
• Option 2: Two separate analyses (this seems confusing)
  
• Option 3: Sum abundance by trap pair and use P/A in analyses
  
• Use relative abundance or PA?
  
• For now I’m planning to do both and compare
  
• Is it fine to use a trap as a sample (with site as a RE)?
  
• I’m worried about independence of samples, but also we only have 9 sites?
  
• Should I remove rare taxa?
  
• I’m thinking of limiting everything to the top 8(?) lizard prey taxa

2. How does lizard diet vary with habitat?

• Visualize with NMDS
  • Point for each lizard gut sample
  • Vectors for veg cover (by life form)
    • These would have to be based on site averages, since we obviously don’t have lizard-specific vegetation data
  • Vector for land use (might need to be binary –> polygon)
• Model with multivariate GLMM (via mvabund)

taxa abundances ~ land use + veg cover (by lifeform) + RE:site

Same questions as above:
- Use relative abundance or PA? (planning to do both and compare)
- Is it fine to use a lizard as a sample (with site as a RE)?
- Should I remove rare taxa? (maybe limit to top 8 prey taxa)

3. Does invert community predict lizard diet?

• Maybe visualize with some sort of correlation matrix?
  • I’m thinking prey items on the x and diet items on the y, color coded by strength of interaction?
• Model with multivariate GLMM (via mvabund)

diet abundances ~ prey abundances + RANDOM:site

• Is this even a thing you can do???
  
• I think it is I just can’t find an example with mvabund yet but haven’t looked very hard
  
• Same as above: Use relative abundance or PA?
  
• I think that

4. What about plants and ants?

• Something simple about herbivory and/or myrmecophagy because they’re especially weird/interesting
• Visualize with
  • Maybe boxplots for land use (if needs to be binary)
• Model with GLMM

herbivory ~ land use + veg cover (by lifeform) + RE:site myrmecophagy ~ land use + veg cover (by lifeform) + RE:site

Other ideas

• Diversity/abundance
• Invert biomass
• Invert functional groups (e.g. hardness, activity level)
• Prey selectivity indices

Qs 4 Colin

• Land use gradient instead of binary?

To do:

• Separate ants using comments
• Flag rare OTUs